A workshop can end with enthusiastic feedback, a group photograph and a folder of slides while leaving everyday work almost unchanged. That does not mean training is useless. It means the event and the business outcome are different things. The distinction becomes especially important when a fashionable topic such as AI encourages organizations to book a session before they have defined the problem participants are supposed to solve.
A stronger training brief starts with a task, the people doing it and the conditions under which they will apply new skills. This article sets out a planning and evaluation method for a small business or a business development service provider. The examples are constructed teaching scenarios rather than claims about completed client projects. The method is designed to make an assignment more testable, not to promise a particular return from any course or training provider.
Connect the current skills conversation with an operating need
The OECD’s 2026 SME digitalisation research identifies skills and implementation challenges alongside growing AI use. It should be read with its sampling limitations in mind and cannot establish the training needs of an individual Ethiopian enterprise. [OECD: Empowering SMEs in the age of AI — 2026 D4SME Survey] A business still has to identify which task it wants people to perform differently.
The ILO’s SIYB programme offers a useful contrast to an event-only view of learning: its public description connects business development stages with training and follow-up support. The generic trainer’s guide also includes needs analysis, preparation and monitoring in the training cycle. These are references for the importance of a learning process, not evidence that any workshop bearing a similar name is ILO-accredited. [ILO: Start and Improve Your Business programme] [ILO: Start and Improve Your Business generic trainer’s guide] Ask the prospective provider to describe the actual approach and evidence behind the proposed assignment.
Define the task before selecting the course title
Replace a broad request such as “train our team in AI” with a description of work. A small consultancy might want staff to prepare a first draft of a client meeting summary while preserving decisions, uncertainties and agreed actions. Another business might need a clearer procedure for responding to standard product enquiries.
These are different learning needs even if both involve an AI tool. The first requires careful interpretation of notes and agreement about what belongs in an official record. The second requires an approved product reference and a route for questions outside that reference. Write down the current task, a typical input and an acceptable output. Then ask what knowledge and practice participants need to move between them. A trainer can respond more precisely to that brief than to a request for a broad introduction that may not fit the participants’ daily responsibilities.
Understand the participants’ starting point
A team can share a job title while having very different experience with the tools used in training. Check access to devices, familiarity with the interface, language needs and the time participants can realistically allocate. These questions should be treated as design inputs rather than reasons to exclude people automatically.
Use a short, relevant starting task where appropriate. For the meeting-summary example, participants could prepare a summary from the same fictional notes before the session. The trainer can then see whether the main difficulty is identifying decisions, using the software or writing clearly for the intended reader. Avoid collecting sensitive client material merely to make the exercise feel realistic. A carefully designed fictional case can test the required skill without exposing confidential information. Explain what the starting task will be used for so that participants understand it as a way to improve the programme rather than a hidden performance assessment.
Make the learning objective observable
An objective such as “understand AI” is difficult to evaluate. A more useful objective describes an action and the standard that makes it acceptable. For example, participants might be expected to produce a meeting summary that separates decisions from proposals, identifies unresolved questions and does not add unsupported commitments.
The business and trainer should agree those criteria before the session. They can then design exercises that expose the relevant mistakes. A polished summary that invents a deadline should not pass merely because the writing is fluent. The objective should also identify what remains a human responsibility. Participants need to know that using a drafting tool does not transfer accountability for a client record. Clear criteria help the trainer focus feedback and help the manager recognize whether the learning has reached the level needed for actual use.
Choose practice that resembles the workplace
A demonstration lets participants watch a skilled person complete a task. Practice lets them encounter the decisions they will need to make independently. Both can have a place, but they should not be mistaken for the same activity. Reserve enough time for participants to try the task, receive feedback and revise their output.
Use realistic complications. The fictional meeting notes could contain conflicting dates, an unfinished discussion and a suggestion that was not approved. Participants should identify the uncertainty rather than allowing a tool to turn every sentence into an agreed action. Introduce a second example after the first has been discussed so that people cannot succeed only by copying the trainer’s exact wording. The aim is transfer of judgement to a new but related task. That requires a more thoughtful exercise than a sequence of screenshots showing where to click.
Give managers a role after the workshop
Staff may learn a skill but return to a workflow that does not allow them to apply it. A manager can help by making approved references available, allocating review time and clarifying which assignments are suitable for initial practice. Without those conditions, participants may either avoid the tool or use it without the checks discussed in training.
Agree a small workplace assignment before the course begins. In the consultancy example, each participant might prepare a draft from a non-sensitive internal meeting, using the same review criteria practiced in the session. A designated reviewer would provide feedback and record common issues. The manager should know how much time this will require. Follow-up is more likely to occur when it is part of the assignment, with a responsible person and a planned review, rather than an optional suggestion at the end of a busy day.
Measure behaviour separately from satisfaction
Participant feedback can show whether the session was clear, accessible and relevant. It cannot establish on its own that workplace performance improved. Use different measures for different questions. Ask about the learning experience, review a task against agreed criteria and later observe whether the relevant practice is being used.
For a hypothetical group, the business might count how many summaries require correction for unsupported actions before and after training. Those observations would need comparable tasks and an explanation of who reviewed them. Do not present an illustrative measurement design as a demonstrated result. If the work changes during the period, describe that limitation. A useful evaluation can report that participants learned a technique while also finding that an incomplete reference process prevents consistent application. That finding points toward a management action rather than a need to repeat the same workshop automatically.
Keep costs visible without inventing a return
The course fee is only part of the commitment. Participants spend time attending, practicing and receiving review. The business may need devices, connectivity, a tool subscription or updated procedures. These are relevant planning inputs even when a precise monetary return cannot be established.
Compare the assignment with the problem it is intended to address. If the objective is more reliable meeting records, look for evidence about record quality and review effort. Do not promise that a short workshop will cause a particular revenue increase. That would require much stronger evidence and a way to separate the effect of training from other changes. A transparent account of the resources used and the behaviour observed gives the owner a more defensible basis for deciding whether to continue, redesign or narrow the programme.
Ask providers for evidence relevant to your brief
A provider’s general experience can be useful, but relevance matters. Ask how the proposed method addresses the participants’ starting point and the target task. Request examples of exercise design, feedback criteria and follow-up arrangements that can be shared without exposing another client’s confidential information.
Check any claimed certification or institutional affiliation through an appropriate source. A programme name, a logo on a presentation or a photograph of a previous event does not establish that the proposed trainer holds a current authorization. This is a verification step rather than an accusation. The provider should also explain the limitations of the assignment. An honest proposal can distinguish what the session will cover, what requires further practice and what depends on the client’s own operating environment. That clarity is more valuable than a promise to transform every part of the business in a single day.
Make the final product a usable working habit
The most useful legacy of training may be a small set of examples, a review checklist and a clear boundary around where the new skill is appropriate. Those materials should be available to participants in a format they can use during real work. Agree who will keep them current as the task or tool changes.
Before booking the next fashionable workshop, write one sentence describing the behaviour you want to see afterwards. Identify a task that can demonstrate it and a person who can review it. Then build the programme around that evidence. A course can still be engaging and ambitious, but its value becomes easier to discuss when the organization can show what people practiced, what changed at work and what further support is needed to make the change last.
